Beyond presence and frequency: a determinant-level framework for interpreting AMR dissemination in One Health surveillance.

Genomic sequencing is now routine in One Health antimicrobial resistance (AMR) surveillance, but reports often stop at presence and frequency and do not distinguish a determinant confined to one genomic background from one recurring across lineages, reservoirs, and reporting periods-patterns that warrant different action. We propose the Determinant Dissemination State (Dg-state), a determinant-level interpretation layer that classifies each resistance determinant as Contained, Emerging, Disseminating, or Abstain. We specify the complete decision rule under a single notation: six bounded evidentiary domains; a composite dissemination score with default weights; an uncertainty-adjusted score that discounts weak genomic context and thin sampling; an explicit sampling-adequacy gate; a two-stage classification separating confined recurrence from genuine cross-reservoir or cross-lineage spread; and a classification-stability measure. As a proof-of-concept, and with an accompanying reference implementation, we apply the framework to a literature-anchored simulated surveillance dataset with known ground truth, where it reproduces the intended labels and behaves robustly. Dg-state does not infer transmission; it provides an auditable, reproducible way to prioritise determinants for escalation, targeted resampling, and higher-resolution genomic investigation.